MiNet: A Convolutional Neural Network for Identifying and Categorising Minerals
This work addresses mineral identification for geologists and field researchers, but it is incremental as it applies an existing deep learning method to a new domain-specific task.
The paper tackled the problem of identifying minerals from hand specimen images by developing a convolutional neural network (CNN) model, achieving an accuracy of 90.75% on a real-world dataset with seven mineral classes.
Identification of minerals in the field is a task that is wrought with many challenges. Traditional approaches are prone to errors where there is no enough experience and expertise. Several existing techniques mainly make use of features of the minerals under a microscope and tend to favour a manual feature extraction pipeline. Deep learning methods can help overcome some of these hurdles and provide simple and effective ways to identify minerals. In this paper, we present an algorithm for identifying minerals from hand specimen images. Using a Convolutional Neural Network (CNN), we develop a single-label image classification model to identify and categorise seven classes of minerals. Experiments conducted using real-world datasets show that the model achieves an accuracy of 90.75%.